Bluenami · Article I · Five Weeks Out v.1
Bluenami Congress v.1 · Nineteen red-state Senate seats and everything else on those ballots · data to Sept. 29, 2026 Bluenami · Article I · Five Weeks Out · v.1 · 1980 and 2026 side by side · data to Sept. 29, 2026
Five weeks before November 3 · second in a dated series after Six Weeks Out

Nearly half a century ago the polls called it too close to predict, and on election night forty-four states went the same way.

Two maps sit side by side here: 1980, when the country went Republican nearly everywhere after a September in which the polls had Reagan a few points ahead and called the race too close to predict, and 2026, as the Senate races stand today. The point, for people who live in nineteen of the twenty-two red states with Senate seats on the ballot this year and for anyone who has written those states off, is that a swing of that size reaches the Senate too, as it did in 1980, and that in 1980 and again in 1994, the two biggest surprises of the last sixty years, it looked smaller from here than it turned out to be on election night.

1980

President · result
Reagan 44 · 489Carter 6 + DC · 49
The Senate the same night, red and blue by winner, black outline where the seat changed party · Republicans +12
Six public polls, Sept. 12–25Reagan +3.5
Five weeks out, ABC/Harris, Oct. 3–6Reagan +4
Mid-October, Gallup, Oct. 17–20Carter +1
Result, Nov. 4Reagan +9.7
SenateR +12 → 53–46–1
HouseR +34 or 35
The six: Gallup, Sept. 12–15, Reagan +4; Washington Post, mid-September, even; CBS/New York Times, mid-September, Carter +3; ABC/Harris, Sept. 22, Reagan +6; AP/NBC, through Sept. 24, Reagan +9; CBS/New York Times, Sept. 23–25, Reagan +5. Simple average, no weighting. Five weeks out, the only national poll fielded that week had Reagan at 43, Carter 39 and John Anderson 14, and Anderson was fading. Two weeks later Gallup had Carter ahead by a point. The race did not open up until the final days, after the Oct. 28 debate, which is part of why the polls kept calling it close.

2026Forecast, not a result

Cook Political Report ratings · unchanged since Sept. 23
Generic ballot, FiftyPlusOne average, Sept. 29D +7.7
Same, Silver Bulletin, likely voters, Sept. 29D +9.6
Trump approval, Quinnipiac, Sept. 24–2734% approve, 61% disapprove
Senate now53–47 · Democrats need net +4
House now218–214–1, 2 vacant · Democrats need net +4, to 218

The 1980 map shows the presidential result because that is what turned the map one color. There is no presidential race in 2026, so the second map opens on the Senate races as the Cook Political Report rates them, which it summed up on Sept. 23 as "a true toss up" for control and has not changed since. The fifteen states with no Senate race this year are striped by their current senators. The outlines mark races that any of the three main raters moved in September. The other two views show a count of Cook's close House districts, rated Sept. 25, and, for the nineteen states only, the governor's race where there is one this year (fourteen of them), with the Senate race elsewhere.

Since Six Weeks Out

What changed in the last week, and what did not

Part one

How big waves looked in September

1964 is the case you could see coming and 2006 looked bigger in September than it turned out; 1980 and 1994 looked smaller than they were, and 2018 came in about as advertised. The small Senate maps show the winning party in each state with a race that year, black outlines where the seat changed party; states with no race are shown in gray. The 2006 and 2018 results are from the Federal Election Commission; 1964, 1980 and most of 1994 are from Wikipedia's per-year pages, which cite the House Clerk's statistics; those were not checked against the originals here.

YearMapSix weeks outWhat happened
1964
President · Johnson 44 + DC, Goldwater 6
Senate · results by state
Gallup September average Johnson +33Johnson +22.6; Democrats +36 or 37 House seats, +2 Senate to 68–32
1980
President · Reagan 44, Carter 6 + DC
Senate · results by state
Public polls Reagan +3.5; Republicans had counted on four or five Senate seats, as Time reported afterwardReagan +9.7; Republicans +12 Senate seats and the majority, nine Democratic senators beaten
1994
Senate · results by state
Gallup September R +4; forecasters expected 25 to 30 Republican House gainsRepublicans +52 to 54 House seats and the majority for the first time in forty years; +8 Senate seats, 52–48 on the night and 53–47 after Shelby switched a day later
2006
Senate · results by state
Gallup early September D +12 among registered voters; one likely-voter reading tiedDemocrats +31 House; +6 Senate seats (Missouri, Montana, Ohio, Pennsylvania, Rhode Island, Virginia), taking both chambers 51–49 with the two independents
2018
Senate · results by state
Trump approval about 40 percent; Cook House toss-ups in Republican seatsDemocrats won 23 of Cook's 28 House toss-ups in Republican-held seats, +40 or 41 House seats; Republicans +2 in the Senate, losing Arizona and Nevada and taking North Dakota, Missouri, Indiana and Florida
2026
Cook ratings, Sept. 23 · governor's race in fourteen of the nineteen states, Senate elsewhere
Generic ballot D +8.1; Trump approval 32–35% in polls fielded Sept. 16–21—

The pattern is not a rule. In 1998, 2002 and 2022 the president's party did better than its September numbers suggested, and 1964 shows a popular president whose party gained. What the three midterm waves above share with 1980 is a sitting president well under fifty percent approval asking voters to treat the election as a verdict on him. Tennessee changed hands twice in 1994, so that map shows seven outlined states for eight seats; Minnesota and Mississippi each held two races in 2018, both won by the same party; Connecticut in 2006 is shown as an independent win, Lieberman's, and not outlined as a change of party. Independents are shown in green.

Part two

How many would have to switch*

Thirty-four days remain. That is time enough to move expectations, the story people tell about a race, and votes themselves, and this section puts the vote part in plain numbers. Each bar is one state's 2024 presidential vote laid out left to right, Harris voters then Trump voters, with a line at the point where the two would be equal. The blue stretch runs from where the Democratic vote ended to that line: it is the number of Trump voters who would have to cross for the state to tip, half the 2024 margin. The figures on the right give it in votes, as a share of Trump's 2024 vote, and as a share of everyone who voted. For scale, the generic ballot has moved about eight points toward Democrats since 2024, and a uniform eight-point swing would tip none of these nineteen states on its own; the closest, Ohio, starts eleven points back.

State2024 vote, left to right · blue is the askThe ask
*If that many switch, the state tips. That is the whole of what these bars say. They do not include turnout, which has fallen in past midterms and fallen unevenly, and the measuring stick is a presidential-year electorate, larger than any midterm's. In Idaho, Nebraska and South Dakota the Senate opponent is an independent rather than a Democrat, so the crossing is to a different name on the ballot. The bars also do not include attempts to interfere with the election, which this year come from the federal government as well as from adversaries outside the country; the courts have already blocked the Postal Service from carrying out one order limiting mail voting. What is here is a bar, set by the last time everyone in these states voted on the same day. Everything else moves it.
Part three

Nineteen red states, office by office

These are nineteen of the twenty-two states where Republicans are defending Senate seats this year, leaving out Iowa, Maine and North Carolina, which were always the main battlegrounds. Across the nineteen, about 105 statewide races are on the November 3 ballot, counting the Senate seats, governors, the other elected executive offices and the supreme court seats chosen statewide, and leaving out Alabama's four appellate seats and the Texas high-court seats, whose candidates are not listed here. At least twelve of those races have no Democrat on the ballot, including the Senate races in Idaho, Nebraska and South Dakota, where an independent is the opposition. Five of the states, Kentucky, Louisiana, Mississippi, Montana and West Virginia, elect no governor this year. Below the Senate and governor races, only one has a public poll from August or September, the Kansas secretary of state race, and that poll is a Democratic one.

Statewide races on the ballot

about 105
Judicial retention votes in Alaska, Florida, Kansas and Oklahoma were not verified and are left out.

With no Democrat

12+
Confirmed. Several more offices have Democratic nominees that could not be found, so the true count may be higher.
LV means likely voters, RV registered voters; (D) marks a poll from a Democratic sponsor. Last column: official checked against a state or county election office list · reported from news guides, Ballotpedia or Wikipedia · not found

Three things here are checked against official sources: Ohio's candidate list (Erie County Board of Elections), Wyoming's retention justices (Wyoming Secretary of State) and Cook's Senate ratings (Cook's own Sept. 23 chart). Rows marked reported have not been matched to the state's certified list. The 2020 Senate margins are approximate. Two points remain unsettled: Sabato's Ohio secretary of state rating, given as Likely R and Leans R in different places, and whether David Leslie (D) is in the Alaska Senate field.

Sources